Autocorrelated Data and Dynamic Systems

Abstract

Dynamic systems are represented by variables that change in time, and are related to their values in the past. Linear time series models provide a framework for fitting dynamic data. The most important feature exploited by these models is called the autocorrelation (or partial autocorrelation) function. These functions can be estimated under certain regularizing conditions, known as stationarity. When the data-generating process is stationary, the estimates of the auto- and partial autocorrelation functions can be used to suggest a model.